Flow Measurement Uncertainty
Using Tracer Gas Dilution Method Eric Harman
CEESI
54043 WCR 37
Nunn CO 80648
970-897-2711
Comparative Flare Gas Flow Measurement Study
bull In 2011 Chevron contracted a blinded study to test
various flow meters used in flare gas measurement
bull The goal was to shed some light on different flare gas
flow measurement technologies
bull Improve
o API-1410 (Measurement of Flow to Flares)
o API-223 (Testing Protocol for Flare Gas Metering)
bull Data presented at the GPA (Gas Producers
Association) April 2013 San Antonio Texas
Thank you Thank you Eric Estrada Targa Resources Steve Baldwin Chevron
Houston Texas USAHouston Texas USA
Comparative Flare Gas Flow Measurement Study
The comparative blinded study ran from 2011
to 2013 and included the following meters
bull USM (4-path Chordal)
bull USM (2-path Diametral)
bull USM (1-path Diametral)
bull USM (1-path Partial Insertion)
bull Optical Flow Meter
bull Tracer Gas Dilution Methodology
Wanted to test Pitot Tube Technology but
time amp money didnrsquot allow it
Comparative Flare Gas Flow Measurement Study
Consider How a Meter Senses the Flow
Comparative Flare Gas Flow Measurement Study
Rules of the Game
Fluid Air
Temperature 70degF (Ambient)
Pressure 12 PSIA (Ambient)
Velocity 1 to 150 FPS (feetsecond)
Pipe size 10rdquo (6rdquo pipe for 4-path chordal USM)
Pipe orient Horizontal
Piping Config Ideal straight-run
Swirling flow after an elbow
In-Plane amp Out-of-Plane
Uncertainty Using Tracer Gas Dilution Method
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
bull In 2011 Chevron contracted a blinded study to test
various flow meters used in flare gas measurement
bull The goal was to shed some light on different flare gas
flow measurement technologies
bull Improve
o API-1410 (Measurement of Flow to Flares)
o API-223 (Testing Protocol for Flare Gas Metering)
bull Data presented at the GPA (Gas Producers
Association) April 2013 San Antonio Texas
Thank you Thank you Eric Estrada Targa Resources Steve Baldwin Chevron
Houston Texas USAHouston Texas USA
Comparative Flare Gas Flow Measurement Study
The comparative blinded study ran from 2011
to 2013 and included the following meters
bull USM (4-path Chordal)
bull USM (2-path Diametral)
bull USM (1-path Diametral)
bull USM (1-path Partial Insertion)
bull Optical Flow Meter
bull Tracer Gas Dilution Methodology
Wanted to test Pitot Tube Technology but
time amp money didnrsquot allow it
Comparative Flare Gas Flow Measurement Study
Consider How a Meter Senses the Flow
Comparative Flare Gas Flow Measurement Study
Rules of the Game
Fluid Air
Temperature 70degF (Ambient)
Pressure 12 PSIA (Ambient)
Velocity 1 to 150 FPS (feetsecond)
Pipe size 10rdquo (6rdquo pipe for 4-path chordal USM)
Pipe orient Horizontal
Piping Config Ideal straight-run
Swirling flow after an elbow
In-Plane amp Out-of-Plane
Uncertainty Using Tracer Gas Dilution Method
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
The comparative blinded study ran from 2011
to 2013 and included the following meters
bull USM (4-path Chordal)
bull USM (2-path Diametral)
bull USM (1-path Diametral)
bull USM (1-path Partial Insertion)
bull Optical Flow Meter
bull Tracer Gas Dilution Methodology
Wanted to test Pitot Tube Technology but
time amp money didnrsquot allow it
Comparative Flare Gas Flow Measurement Study
Consider How a Meter Senses the Flow
Comparative Flare Gas Flow Measurement Study
Rules of the Game
Fluid Air
Temperature 70degF (Ambient)
Pressure 12 PSIA (Ambient)
Velocity 1 to 150 FPS (feetsecond)
Pipe size 10rdquo (6rdquo pipe for 4-path chordal USM)
Pipe orient Horizontal
Piping Config Ideal straight-run
Swirling flow after an elbow
In-Plane amp Out-of-Plane
Uncertainty Using Tracer Gas Dilution Method
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
Consider How a Meter Senses the Flow
Comparative Flare Gas Flow Measurement Study
Rules of the Game
Fluid Air
Temperature 70degF (Ambient)
Pressure 12 PSIA (Ambient)
Velocity 1 to 150 FPS (feetsecond)
Pipe size 10rdquo (6rdquo pipe for 4-path chordal USM)
Pipe orient Horizontal
Piping Config Ideal straight-run
Swirling flow after an elbow
In-Plane amp Out-of-Plane
Uncertainty Using Tracer Gas Dilution Method
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
Rules of the Game
Fluid Air
Temperature 70degF (Ambient)
Pressure 12 PSIA (Ambient)
Velocity 1 to 150 FPS (feetsecond)
Pipe size 10rdquo (6rdquo pipe for 4-path chordal USM)
Pipe orient Horizontal
Piping Config Ideal straight-run
Swirling flow after an elbow
In-Plane amp Out-of-Plane
Uncertainty Using Tracer Gas Dilution Method
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
-250
-200
-150
-100
-50
00
50
100
150
200
250
05 50 500
Pe
rce
nt
Erro
r
Velocity (ftsec)
Comparative Blinded Flare Gas Flow Measurement StudyTracer Gas Dilution Method
CEESI 10 Pipe May 7-8 2013 CE-18143
Tracer Gas Data
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
Consider How Tracer Gas Dilution
Measures Flowrate
Itrsquos Magic
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really works
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
How Tracer Gas Dilution Method
Really Works
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
Flow Rate Equation
From ASTM E2029
Where
119917119932 = Upstream mass flow rate
119914119920 = Injection stream concentration of tracer gasdagger
119914119915 = Downstream concentration of tracer gasdagger
119914119932 = Upstream concentration of tracer gasdagger
119917119920 = Injection mass flow rate
dagger All concentrations are mass concentrations
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
Uncertainty Equation
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Bias Errors
SF6 toxicity threshold = 1000 ppm
ACGIH recommends 110th toxicity threshold therefore
119914119920max= 00001
= min flowrate = 1 ftsec in a 10rdquo pipe = 0034 lbssec119917119950119946119951 At amp a max SF6 conc of 00001 119917119920 = 0012 lbshr 119917119950119946119951 This extremely small flow rate can be achieved by diluting
the tracer gas which adds to uncertainty of 119917119920 alternately an
extremely small thermal mass meter could be used with
pure SF6
The term is estimated to be (004)2
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
120784∆119914119915 120784+ ∆119914119932The term is related to the gas analyzer 119914119915minus119914119920
uncertainty which depends on the calibration gas
and the analyzerrsquos ability to match the calibration gas
value Typical calibration gas uncertainty for SF6
concentrations of 1ppm to 100 ppt = 2 Estimated
analyzer uncertainty to match calibration gas = 05
120784∆119914119915 120784+ ∆119914119932The term is estimated to be (002)2
119914119915minus119914119920
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
More Uncertainty Calculations
Random error due to the thermal mass meter amp the
analyzer
Multiple samples were taken to characterize the
random error in the thermal-mass meter amp the
analyzer the 2σ uncertainty was 506
120655119931119917
119917 = 120782120786 120784 + 120782120784 120784 + 120784 120782120787 120784= 83
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
A Few Pictures
Test Facility amp Test Piping
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
Tracer Gas Dilution Installation
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
Results
Question
From CEESI Testing Trace Gas
Dilution Method Uncertainty
was plusmn6-10 Is this reasonable
Answer
Yes uncertainty calculation
revealed an 83 uncertainty at
2-sigma
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Uncertainty Using Tracer Gas Dilution Method
Trace Gas Dilution Observations
bull The worse the straight-run the better the mixing
bull A large error due to the injection flow rate is possible
bull Consider the calibration gas uncertainty in the
uncertainty calculations
bull Because concentration sampling is a discrete single
point measurement large random errors are likely so
many samples are required to reduce random error
bull How well the tracer gas mixes is hard to characterize
Tracer Gas Dilution is labor
intensive but may be a viable
alternative when other methods
are not possible
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you
Comparative Flare Gas Flow Measurement Study
Comments amp Questions
Thank you